swin-tiny-patch4-window7-224-finetuned-eurosat

This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.3937
• Accuracy: 0.8583

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6866	0.99	37	0.6202	0.6742
0.5892	2.0	75	0.5467	0.72
0.5195	2.99	112	0.4933	0.7483
0.4322	4.0	150	0.4787	0.765
0.3712	4.99	187	0.3829	0.8208
0.3162	6.0	225	0.3960	0.8133
0.3082	6.99	262	0.3591	0.8392
0.3038	8.0	300	0.3274	0.8467
0.2794	8.99	337	0.3533	0.8433
0.2596	10.0	375	0.3766	0.8258
0.2369	10.99	412	0.3392	0.8575
0.2503	12.0	450	0.3198	0.8625
0.2009	12.99	487	0.3438	0.8625
0.2195	14.0	525	0.3234	0.8617
0.2025	14.99	562	0.3758	0.855
0.1879	16.0	600	0.3909	0.8408
0.18	16.99	637	0.3642	0.8617
0.1545	18.0	675	0.3948	0.8567
0.171	18.99	712	0.3889	0.8525
0.1667	20.0	750	0.3883	0.8625
0.163	20.99	787	0.3743	0.8575
0.1682	22.0	825	0.3739	0.8592
0.1611	22.99	862	0.3623	0.8742
0.1348	24.0	900	0.3806	0.8592
0.1366	24.99	937	0.3849	0.865
0.1418	26.0	975	0.4049	0.8558
0.1096	26.99	1012	0.3849	0.8608
0.1347	28.0	1050	0.3926	0.8592
0.137	28.99	1087	0.3938	0.8592
0.1312	29.6	1110	0.3937	0.8583

Framework versions

• Transformers 4.30.1
• Pytorch 2.1.0+cu121
• Datasets 2.15.0
• Tokenizers 0.13.3